I don't know how this would be done because the reflection that generated the weight also needs to be known. So one wouldn't know what to keep and what not to keep with only having a list of the weights. Could you please explain a bit more what you mean?
Justin On 11 September 2010 20:57, Raul Miller <[email protected]> wrote: > On Sat, Sep 11, 2010 at 10:43 AM, Raul Miller <[email protected]> wrote: >> Oh, right, i forgot about that. >> >> You would need to modify keepv to also eliminate rows which only contain >> zeros. > > Specifically, I think you can generate all the needed weights (as well > as a lot of bogus ones) using > > currentLevel- currentLevel*"1 2 A > > And you can of course combine the first two dimensions > using ,/ (though you need those dimensions for some of > the tests, so you should delay that operation until after > you have completed all the tests for validity). > > That said, I am a little over my head here, and I am having > trouble refactoring the tests so they are meaningful using > this representation of the results. > > -- > Raul > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
